TEACHING NOTE: BP-256TN

DATE: 08/31/05

TEACHING NOTE FOR INTEL CORPORATION (A)

THE DRAM DECISION
INTRODUCTION
This case was the first outcome of the research and teaching collaboration between Andy Grove
and myself, which gave rise to the course on Strategy and Action in the Information Processing
Industry at Stanford Business School, and on which Strategic Dynamics: Concepts and Cases is
based. It is the most complex case of the entire textbook and requires instructors to familiarize
themselves with a greater amount of technical details than any of the other cases in the book. On
the other hand, as it describes one of the 20 most interesting (and important) corporate
transformations of the 20th century (Fortune Magazine, June 27, 2005). It also illustrates the
three key themes of the book and all its conceptual frameworks, and it is a classic case of what
we have called P-independent change. So, while some instructors might decide to skip this
opening case because of its technical complexity, those who are willing to familiarize themselves
with enough of technical detail to engage in a strategic rather than technical discussion will
gain tremendous insight in some of the fundamental strategic dynamics of high-technology
industries.
In order to help instructors willing to make this investment, this teaching note goes into great
technical detail and references all relevant exhibits and case facts necessary to understand at the
strategic level the differences between the three key technological competences - circuit design,
process technology, and large-scale precision manufacturing - necessary to understand how the
DRAM business changed from a specialty business to a commodity business, and how and why
Intel failed to defend its strategic position in the evolving DRAM industry. I extend a special
thanks to George Cogan, Stanford MBA 1989 whose thorough understanding of semiconductor
technology was indispensable for writing parts of this teaching note.
CASE OVERVIEW
The case focuses on the decision Intels top management faces at the end of 1984: whether to
exit the DRAM (dynamic random access memory) business. It serves as a vehicle to discuss the
This note was prepared by Robert A. Burgelman for the sole purpose of aiding classroom instructors in the use of Intel
Corporation (A): The DRAM Decision (Stanford Business School Case). It provides analysis and questions that are intended to
present alternative approaches to deepening students comprehension of business issues and energizing classroom discussion.
Copyright 2005 by the Board of Trustees of Leland Stanford Junior University. No part of this publication may be reproduced,
stored in a retrieval system, used in a spreadsheet, or transmitted in any form or by any meanselectronic, mechanical,
photocopying, recording, or otherwisewithout the permission of Stanford University.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 2

evolution of Intels distinctive technological competencies and of the critical linkages between
its corporate, technology, marketing, and manufacturing strategies. The case also provides a rich
context in which to examine the effect of apparent industry structural changes on the relative
importance of different technological competencies and on the linkages between corporate,
technology, marketing, and manufacturing strategies. It is a classic example of P-independent
change: Industry changes over which the focal company (in this case Intel) had virtually no
control. As a result, while Intel was the first successful mover with DRAM products it finds itself
with an inconsequential market segment share at the end of 1984. While some at the top of the
organization still believes that DRAMs are critical both from market and technology viewpoints,
DRAMs have actually played a minor role in Intels product mix in the past five years. Logic
products based on microprocessor technology have replaced memory products as Intels core
business. The case leads to discussions about the implications of organizational beliefs on
technology and manufacturing strategy and top managements role in evaluating those beliefs
during changing times. Of particular importance in this discussion is the distinction between the
product-market and technological competence aspects of the exit decision that are strongly
intertwined in top managements assessment of the strategic situation. The case also leads to
discussions about the role of the internal selection environment in a firms evolution. Finally, the
case leads to a discussion of the process of strategic renewal at Intel, and the role of top
managements capacity for strategic recognition in that process.
MAJOR THEMES
1. Distinctive technological competencies and the linkages between technology, manufacturing,
marketing, and corporate strategies.
2. P-independent change: The impact of a changing competitive environment and a rapidly
maturing industry on the basis of competition, and on a firms strategic position in the industry
and the relative importance of its distinctive competencies and the linkages among them.
3. The tradeoff between scope and depth in core technologies and the difference between firmand industry-level technology drivers.
4. The implications for corporate strategy of tensions between commodity and proprietary
businesses, and the divergence between stated strategy and strategic action that a shift from
specialty business to commodity business may cause.
5. The difference in strategic relevance between cyclical business fluctuations and structural
industry changes where structural changes may include a shift in equipment supplier function
and the entrance of a new group of competitors with a different model for competitive behavior.
6. The impact of organizational culture and management control systems the internal selection
environment in shaping corporate strategy and the implications for top management when
stated corporate strategies and strategic action diverge.
SUGGESTED ASSIGNMENT QUESTIONS
1. What should Intel do regarding the DRAM business in November 1984? Why? Evaluate the
four options laid out at the end of the case to support your recommendation.
2. At the end of 1984, how important is the DRAM business for Intel going forward? What did
different top-level executives at the time say about this? Why?

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 3

3. Why did Intel lose its leadership position in the DRAM industry? What has Intels technology
strategy been with respect to DRAM? What does this say about Intels generic strategy and the
image it has of itself as a company? How has the organization reacted to the evolution of the
DRAM business? Why could the organization do so? What could Intel have done differently?
What would be the implications if it had done so?
4. How did Intel get into the microprocessor business? How important is this business for Intel at
the end of 1984?
5. What is the strategic conundrum that Intel top management has faced since 1982, and how can
it be resolved by Dr. Grove at the end of 1984 in order to drive Intel in a viable new strategic
direction?
DISCUSSION OF MAJOR THEMES AND SUPPORTING ANALYSIS
1.

Distinctive Technological Competencies and the Linkages between Technology,

Manufacturing, Marketing, and Corporate Strategies

It is important to realize that there are basically three major technological competencies
associated with semiconductors: (1) circuit design (can we design it?), (2) process technology
(can we make it?), and (3) large scale precision manufacturing (can we make it in large volumes
with high yields?). Intels initial distinctive technological competencies were in design and,
especially, process technology. Intels corporate strategy has evolved since 1972. In the early
days, Intels strategy was to develop, maintain, and exploit technical leadership by focusing on
processing leadership and leading edge products. Its black magic semiconductor processes
provided a clear competitive advantage, which led to a cost advantage over older technologies.
The technology strategy was to develop new semiconductor processes before competitors; the
marketing strategy was to use a cost advantage to dominate existing technologies. Note that
Grove and Moore both come from process technology backgrounds. Moore was a Chemistry and
Physics Ph.D., and Grove worked for him at Fairchild.
By the end of 1984, the strategy has changed on two levels: first, the process technology
capability of competitors has been leveled through mobility of engineers and participation of
equipment suppliers in technology development; second, microprocessors and related logic
products have become important (see case Exhibit 7). Microprocessors and DRAMs differ from
each other both from technology and marketing strategy perspectives.
From a marketing standpoint, DRAMs have become a commodity product The DRAM business
went through a rapid evolution from specialty to commodity product Different manufacturers
DRAM offerings are now compatible, and new generations are rapidly adopted by the
marketplace.
Intels DRAM marketing strategy has been to enter and exit each new generation early to
maximize margins and avoid severe price competition (see case Exhibit 3 for a time line of
DRAM introductions and case Exhibit 4 for Intels market share versus price in various
generations). Intel places a high value on being first with new product features. Its current

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 4

strategy in DRAMs appears to be an attempt to differentiate. The organization uses its success
in the one-power-supply 16K DRAM to justify the differentiation strategy (see case Exhibit 4).
While the 16K device first appears to be a successful product, a close examination of case
Exhibit 4 reveals its overall volume is rather small in comparison with other 16K DRAMs.
Intels technology strategy for DRAMs appears to match its differentiation marketing strategy.
Intel attempts to be the first at new process and design variations. There is an undercurrent of
preference for engineering/processing elegance at Intel versus the brute force approach of
some US and Japanese competitors. For example, Intel was the first company to attempt to use
redundancy, whereas, TI chose to further reduce defect levels.
Intel has historically used DRAMs as a technology driver. A technology driver leads a
companys development of process improvements. DRAMs make a good technology driver for
the industry for two main reasons: as a product with a relatively straightforward and predictable
design process, it provides the ability to design product and process simultaneously and to
achieve linewidth reduction (smaller geometries help increase manufacturing yields and improve
chip performance); as a high volume commodity product with rapid market acceptance for each
new generation, it provides ample and timely opportunity for learning through manufacturing.
While Intels strategy of being first with new DRAM process technology fits the notion of
technology driver, its position as a low volume player in 1985 does not.
There is an interesting twist to Intels leadership strategy in DRAM technology. While the
marketing strategy for the 256K and 1 MB generations is consistent with a desire to differentiate
(this time through the application of CMOS processing), the technology strategy is becoming one
of optimization rather than new technology leadership overall. Excellence in one key aspect of
processing technology (thin dielectric) has led Intel to stay with proven processing techniques on
the 1 MB generation while the Japanese take the lead on trench-etched capacitors. Chou notes
that Intel chose to focus on optimizing line width rather than to introduce the latest advances in
trench capacitors (e.g., by then we will be able to take advantage of [competitors] learning).
In this narrow sense, Intels ability to remain the technological leader seems increasingly limited
even though it is still first to implement CMOS, and continues to rely on its marketing strategy of
differentiation.
This subtle change in technology strategy can lead to a discussion of the technology choices that
Intel has made during each successive DRAM generation. The timeline in case Exhibit 3 shows
that a new generation of DRAMs emerges regularly at 3-year intervals. Intel apparently chose to
get off the industry trend line at the 16K level by devoting scarce resources to its third 16K
version (with a single power supply). The case presents the idea that Intel managers simply
misjudged the lifetime of the 16K generation. The decision to focus resources on the third 16K
offering can also be seen as a consequence of the linkage between a technology strategy which
encourages new processes and a differentiation strategy which encourages unique offerings.
The linkage of Intels technology and manufacturing strategies is also critical. Intels recent
failure in the DRAM business can be attributed to a lack of focus on manufacturing, particularly
if we use the 64K generation as the first major failure. The decision to use redundancy
suggests that defect levels in manufacturing were higher at Intel than at TI. Intels inferior

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 5

manufacturing capability provided the rationale for going ahead with redundancy at an early
stage. (Note that redundancy eventually became the industry standard at the 1 MB level, but Intel
was two generations ahead of the rest of the industry.)
The rise of the microprocessor and other logic products businesses has dramatically changed the
way that Intel perceives the marketing function. The microprocessor design win is very
different from the memory design win because many more factors must be coordinated to
cultivate a microprocessor customer. In microprocessors, the customer makes a long-term
decision about a chip architecture and a company; in memories, the customer only makes a short
term decision, because other chip companies are certain to enter with pin-compatible products in
all but the smallest niche markets.
The CRUSH and CHECKMATE programs illustrate Intels realization of the long-term
implications of its competitors field success in microprocessors. Because the mutual
commitment between Intel and its customers is greater, the microprocessor product line forces
Intel to focus on product differentiation, customer training and development aids, and upward
compatibility. Microprocessors offer Intel the possibility of locking in customers through high
software switching costs.
Because Intels microprocessor customers must develop their own unique software before
placing large orders, the demand for a new microprocessor typically ramps up much more slowly
than for a new memory product (see case Exhibit 4 to compare DRAM unit growth for any
particular generation with microprocessor unit growth for the 16-bit segment). This has
implications for manufacturing and technology strategy.
From a technical standpoint, DRAMs require state-of-the-art process technology, whereas
microprocessors require design excellence. Note that it would be significantly more difficult to
use logic products as technology drivers because the design cycle is long and unpredictable, and
demand typically ramps up slowly and remains relatively low on a per unit basis.
Gordon Moore suggests a key linkage between technology development and the marketing
function when he says that designers like to see their products in high volume. Intel employed
the strategy of using SRAMs for microprocessor process development. The failure of the SRAM
product line can be traced to this subjugation of SRAMs in the designers minds.
Another key linkage between technology development and manufacturing comes from Intels
historical treatment of R&D. Intels philosophy of physically integrating R&D with
manufacturing comes from Moores experience at Fairchild. That philosophy results in the
companys ability to make rapid incremental process changes. While the system has advantages
in facilitating implementation of new processes, it may also have had some drawbacks,
particularly in the area of process proliferation. Moore suggests that process proliferation has to
be checked periodically. Although environmental changes have forced the company to modify its
original blueprint of keeping technology development and manufacturing facilities common, the
three technology groups still use the same equipment as manufacturing facilities.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

2.

p. 6

P-independent Change: The Impact of a Changing Competitive Environment and a

Rapidly Maturing Industry on the Relative Importance of Technical Competencies
and the Linkages with Other Functional Competencies

The entrance of Japanese competitors into the DRAM marketplace has clearly had a profound
impact on Intels leadership strategy. Through excellence in manufacturing (optimization of
existing processes versus creation of new processes) and commitment to the DRAM business,
Japanese manufacturers have emerged as technology leaders (see case Exhibit 3 for DRAM
product introduction timeline Japanese companies dominate after the 16K generation).
As prices have declined sharply in the DRAM marketplace, Intel has reacted by shifting its
capital allocation away from DRAMs (see facilities built since 1980 in case Exhibit 5; they
include Logic, EPROM, and microcontrollers). Additionally, in the monthly allocation of
manufacturing resources, Intel has attempted to maximize margin-per-wafer-start by stopping
all production of DRAMs in Chandler so that the only remaining DRAM plant is in Oregon.
Several of the linkages described above have come under pressure due to the rapidly changing
environment. In particular, the mutual dependence between Intel and its customers has grown
dramatically as a result of the microprocessor business. Marketing appears to be growing in
relative strength within the organization (from case Exhibit 1: SG&A as a percentage of sales
grew from 16% in 1976 to 19% in 1984 while sales grew seven fold). In addition, Intel appears
to have increased its focus on protecting intellectual property through lawsuits.
On the negative side, Intels manufacturing resource allocation response has resulted in a lack
of commitment to the DRAM business. Without a manufacturing base in DRAMs, Intel cannot
hope to lead DRAM technology development in the future. The DRAMs ability to serve its
technology driver role may also be partly compromised by lack of manufacturing allocation. On
the positive side, Intel has lost far less than TI in the past year.
At the same time, since the R&D allocation is not made on the basis of revenues, Intel still has a
substantial R&D commitment in DRAM, which does lead to advanced process knowledge.
However, given Moores quote on the needs of designers, the DRAM group must realize
eventually that they are not getting the manufacturing capital allocation needed to survive in the
long run.
A critical change in the linkage between technology development and product innovation has
resulted from the companys growth and the increasing complexity of semiconductor processing.
In Intels early days, several products with unforeseen markets succeeded as a result of the
efforts of individual entrepreneurs, the closeness of R&D to manufacturing, and the ability of
one individual to marshal a critical mass of resources. The case highlights the examples of
EPROM and microprocessor development and their unanticipated success stories. Today the
situation at Intel is different. Moore points out that Intel requires a substantial justification to
start a new product. In addition, one person can no longer shepherd a product through all of its
development stages. A major new product may require the coordination of hundreds of people.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

3.

p. 7

The Tradeoffs Between Scope and Depth in Technologies and the Difference Between
Firm-Level and Industry-Level Drivers

Intel is very focused in comparison to its competitors (see case Exhibit 9 for data on market
share in various semiconductor segments for several competitors). In the past 10 years, the case
suggests that Intel has attempted to narrow its focus; for example, the EEPROM line was
recently eliminated. The decision facing management at the end of 1984 boils down to whether
or not to undertake a further reduction in the scope of Intels distinctive competencies. Since the
actual technology resides within the engineering teams and is embodied in their complex
relationship with the fabrication equipment, the decision to exit the DRAM business will not be
easily reversed.
Intels narrowing focus can be seen as paralleling the maturation of the semiconductor industry.
In a framework of technological evolution, a broad scope of core competencies broadens the
firms flexibility in adapting to and creating change. For example, the combination of EPROM
and logic technologies on one chip led to a new segment of microcontrollers. Adaptability is
most critical in the early stages of an industrys evolution when change is most rapid and least
predictable. In the early days, Gordon Moore and Robert Noyce chose to develop both a bipolar
and a MOS process, because they could not see which direction the industry would take. In 1984,
the MOS share of the digital market is over 80% while bipolar represents about 16%, and Intel
has all but exited the bipolar market (see case Exhibit 9).
Even though scope may be most critical in the early stages, a firms adaptability remains
important even in a mature industry, particularly when different competencies can have
unforeseen synergies or when alternative technologies represent potential threats (for example,
the threat of semiconductor components to tube manufacturers). Intels view of competitive
technologies is that they pose a threat only in the distant future. The question of future synergies
between Intels core technologies is more uncertain. Intel may limit some potential future options
by eliminating its DRAM operation.
For example, the trend in microprocessor evolution has been to incorporate more and more of the
peripheral circuitry. Should processing technology advance to a stage where it makes economic
sense to integrate DRAM and microprocessor technologies, the firm with mastery in both
processes would presumably have an advantage. Thus the impact of Intels potential exit from
the DRAM business must be viewed in terms of effects on its distinctive technological
competencies as well as its product market position.
The advantages of focus are clear. The firm that focuses trades flexibility for efficiency. The case
suggests that Japanese success in memories has resulted from a strategy of focus. While Intel
devotes tremendous resources to microprocessor design, Japanese manufacturers have (at least
until recently) licensed their microprocessor designs and focused on manufacturing. Should Intel
decide to drop the DRAM line, it will free corporate resources to strengthen the microprocessor
and EPROM lines. Some evidence in the case suggests that further focusing would allow a
needed realignment of development resources.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 8

Ron Smiths group is in charge of both the SRAM and the microprocessor process technology
development. In terms of revenue, the microprocessor business is Intels most important in 1984
(see case Exhibit 7 for share of revenue). Smith made the strategic decision to discontinue work
on the 4-transistor SRAM process technology, which meant that Intel would not be able to
compete in the high-volume (commodity) segment. This choice revealed that key middle
managers now believed that microprocessors were more important than memories for Intels
future. In contrast, Chous group is the best corporate resource for technology development.
(Note also that Chous group led to the failure in the 64K DRAM generation; however, this
failure can be better seen as the result of other weaknesses.) DRAMs are the least important
segment in terms of revenue, and yet command the most effective development team.
Elimination of DRAM development would allow a redeployment so that the most important
segment could have the best team. However, a premature attempt at redeployment might
potentially lead to Chou and his team to leave Intel with attendant loss of the companys most
advanced process technology capability.
An interesting question raised by the case concerns the difference between industry and firmlevel learning. Intel must assess whether or not it can remain near the leading edge of industry
learning without using the DRAM as a technology driver. The conventional industry wisdom
asserts that DRAMs are uniquely suited to developing process improvements for reasons
enumerated earlier. At the firm level, however, other products may serve as technology drivers.
Intel management appears to view the DRAM as its technology driver. That view is stated in the
case and confirmed by the description of the DRAM technology development group: the group is
working on the smallest geometries (1pm versus 1.5i.tm) and they are the best technology
development team at Intel.
In spite of managements comments, Intel has not actually used DRAMs as a technology driver
in the complete sense since 1980. Even though DRAM may still be Intels highest unit volume
product, DRAMs are not as high a volume product for Intel that they are to the industry as a
whole (see case Exhibit 5: only one of 8 plants are in DRAMs; see case Exhibits 2 and 4:
DRAMs represent between 3% and 6% of sales in 1980 through 1984). From a manufacturing
perspective, Intels DRAM differentiation strategy and the forces which act to limit the DRAM
manufacturing base are inconsistent with the classical notion of a technology driver. Intel
appears to be using other components to develop its new technologies. For example, stepper
alignment technology was developed simultaneously on EPROMs and DRAMS. EPROMs are
produced in 5 of 8 plants (see case Exhibit 8) and represent about 15% of total revenue (from
case Exhibit 7 or calculated directly at 12.9% from case Exhibit 9).
It appears that for the industry, DRAMs remain the technology driver by virtue of their volume.
The market size by segment data in Exhibit 9 coupled with the pricing data in Exhibit 4 allow a
rough estimate of the importance of DRAMs in terms of total number of chips produced in the
world. DRAMs represent about 20% (3.6/18.1) of the dollar volume of digital integrated circuits.
Since DRAMs are less expensive than other chips, this represents a lower bound on their
importance on a unit basis. The case writer estimates that on a unit basis, DRAM5 alone
represent over 35% of the worlds output of integrated circuits in 1984.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 9

4.
The Implications for Corporate Strategy of Tensions Between Commodity and
Specialty Businesses
As described above, the requirements for the DRAM (and EPROM) segment and the Logic
segment are different from manufacturing, marketing, and technology strategy perspectives. The
key questions for Intel are: can one company excel in both areas, and can a company survive in
the long term without DRAMs, the high-volume product?
Several arguments point to the difficulty that one firm has in supporting both commodity and
differentiated products. While allocation of capacity to the most profitable products appears
rational, it reduces Intels ability to compete in DRAMs. Intels capacity allocation system works
against any attempt to build both commodity and differentiated products, since the margin-perwafer-start is always less for commodity products (see also case Exhibit 8). When capacity
allocation is reduced, learning on DRAMs is reduced, and manufacturing costs fall behind those
of the focused competitor. One Intel employee referred to this phenomenon as a death spiral,
and suggested that the same thing that happened to DRAMs was happening now to EPROMs.
Intel has attempted to resolve the internal organizational conflict by creating niche markets for
its commodity products. In effect, Intel chooses not to view the DRAM as a commodity product.
The technology development allocation is in marked contrast to the allocation for manufacturing.
Grove and Moore allocate resources for technology development based on their judgment. As a
result, there is less than perfect correlation of level of development effort with product
profitability. If learning from manufacturing experience is as important as it appears, success in
DRAMs would require a similar allocation procedure for manufacturing resources. In order to be
successful in DRAMs, having a top-notch research effort is not sufficient: a firm must have a
commitment to the business.
Intel hired Dean Toombs from TI expressly to run the DRAM business (see case Exhibit 1). This
can be seen as an attempt on Intels part to capture some of the discipline and design-formanufacture for which TI is known. Within the organizational context of Intel, Toombs takes on
the ironic task of positioning Intels DRAM as a niche product at twice the current market
price. The data of the case, however, indicate that there is no viable niche in DRAM: Intel does
not get the expected 2x price premium and the correlation between the price in the high-end
segment (the putative niche) and the price in the standard market (x) seems to be high.
5.
The Difference in Strategic Relevance Between Cyclical Business Fluctuations and
Structural Industry Changes
Intel management must make a judgment about whether the current semiconductor recession
results from normal cyclical processes (semiconductor industry is very capital intensive and is
located far down the food chain), or from a fundamental structural change in the industry.
There is some evidence in the case that the participation of the Japanese manufacturers and the
evolution of the role of equipment suppliers represents a significant structural change.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 10

The DRAM market segment has undergone a significant structural change with the entrance of
Japanese competitors. Since their entrance in 1980, the five largest Japanese manufacturers have
captured 47% of the combined world DRAM and SRAM markets of nearly $5 billion in 1984
(see case Exhibit 9). Japanese manufacturers have demonstrated that manufacturing leadership
and commitment to the business can lead to dominance.
The issue of structural change can lead to a rich discussion of the implications of competing with
a group that has a different model for competitive behavior. While the case does not provide a
great deal of data on Japanese firms, several factors lead to a different Japanese model for
competitive behavior: a coherent national industrial policy; a low cost of capital; and a long-term
perspective. Semiconductors have been targeted as a strategic industry, and the government
encourages companies to develop capabilities. DRAMs are viewed not only as an end in
themselves, but as a means to develop other products such as supercomputers. Japanese
companies also benefit from patient capital; a high savings rate leads to a low capital cost, and
the structure of share ownership leads to a long term view of the world.
As the semiconductor industry matures and manufacturing becomes more complex and capital
intensive, equipment suppliers have begun to play a more important role in technology
development. The case presents evidence that the locus of process innovation is gradually
shifting from the chip companies to the equipment suppliers. The consequence is that
competitors can now acquire a basic level of technology from equipment suppliers. The strategic
advantage of process leadership is diminished as equipment suppliers act to transfer information
between competitors.
Intels traditional strategy of staying ahead of the experience curve using process technology
must be reassessed in light of the change in equipment supplier role. The new structure leads to a
more rapid transfer of processing knowledge between competitors and between countries. In one
sense, Intel has already reacted to this structural change by developing a new set of distinctive
technical competences related to chip design. The design capabilities represent a competitive
advantage, which is less likely to diffuse through supplier relationships.
Other factors suggesting a structural industry change are the emergence of a dominant design
for DRAMs as of the 4K-generation, which made it more difficult to appropriate intellectual
value added, and the fact that customers could easily switch between DRAM suppliers.
Whether or not there has been a structural change in the industry, the recession which is expected
to begin in 1985 will likely result in casualties (see quarterly data in case Exhibit 1). Even if the
1985 recession is another in a series of normal cyclical downturns, the result may be a
structural change in the industry. Should several US manufacturers leave the DRAM business
(which appears possible), they could possibly leave an undersupplied market for those who
remain.
Intel managements decision on the DRAM issue will hinge on its vision of the future for the
semiconductor industry. Several full-line suppliers dominate the current industry; however,
there are some indications that each full-line supplier has a strong position in one or two
particular technologies and a weaker position in others (see case Exhibit 9). For example, TI and

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 11

several Japanese firms have a strong position in DRAMs but are not known for their
microprocessor technology (design as opposed to manufacture). Intel and Motorola have strong
microprocessors, but relatively weak memory products (Motorola is weak in EPROMs and Intel
is weak in DRAMs). Hitachi has a strong DRAM product line (see case Exhibit 9), but has yet to
develop a proprietary microprocessor. While this evidence is anecdotal, it suggests the possibility
that the industry may develop into several segments, each one dominated by different companies.
6.
The Divergence Between Stated Strategy and Strategic Action and the Implications
for Top Management
Andy Grove said that if you want to know an organizations strategy, dont ask, observe. This
appears to be the situation at Intel in 1985. The overall case text frames the DRAM decision with
the organizations conventional wisdom that DRAMs provide a key technology and
manufacturing driver as well as a critical element of Intels product mix. On the other hand, the
evidence in the exhibits show that Intel has been effectively out of the DRAM business since
1980 (see case Exhibit 5 only one of 8 plants are making DRAMs; see also case Exhibits 2
and 4 DRAMs represent between 3% and 6% of sales in 1980 through 1984).
While Intel views itself as the memory company, it has gradually evolved into the
microprocessor company. The evolution appears to have been the result of organizational forces
the internal selection environment and the interaction of those forces with the external
environment. The role of top management in the change process must be reexamined in light of
the organic process which appears to have led to Intels adaptive behavior.
According to Grove, top management must continually reassess the organizations conventional
wisdom and attempt to question self-evident truths. For example, Intel managers claim that
DRAMs are a technology driver. Similarly, the marketing group claims they need DRAMs to
succeed in selling (same quote). Top managements role becomes one of recognizing
inconsistencies in articulated and actual strategy and in ratifying and championing the firms
actual strategy.
TEACHING THE CASE (BASED ON THE ASSIGNMENT QUESTIONS)
1. Discuss the options. Ask the students to write down their decision on a piece of paper.
Get a tally of the vote by show of hand (first side board - 20 minutes).

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

Advantages
License Technology Maintain technological competence
Maintain future options and full
product line
Still first to market
Niche Player

Regain leadership

Exit

Maintain technological competence

and acceptable ROI(?)
Still first to market
All advantages of licensing
Learn to compete in commodity
Focus on manufacturing
competitiveness
Affect Japanese cash flow
Immediate increase in margins
Increased capacity available
Reduce confusion
Increase strategic focus (i386)

p. 12

Disadvantages
Provide competitors access to distribution channels
Koreans have no experience
Opportunity costs
May not be possible without manufacturing
- Still not focused
Not clear there is a defensible niche
Has not worked so far
Huge investment
Too late
Opportunity cost too high
Cannot be both specialty and
commodity producer
Lose technology driver
No full product line
Reputation: quit
Give up legacy and biggest market
Lose Sunlin Chou?
Dont learn to compete
Japanese increase in cash flow
EEPROM?

Note: The instructor should ask the class to vote again on the options. Is the voting different
from the one at the outset of the class? Ask one or two people who changed their vote to explain
why.

2. At the end of 1984, how important is DRAM for Intel going forward? What did different
executives at the time say about this? Why? (Middle board - 15 minutes)
Note: Questions 1 and 2 overlap somewhat. It may be helpful to keep track of the arguments that
reflect beliefs and facts while discussing the options. This requires using the first sideboard and
the first middle board simultaneously.
Beliefs
Technology Driver

>

Facts
Only 3% SOM versus 15% for Hitachi; but DRAM process technology
group still leads in linewidth reduction
Late in 64K: 1.7% SOM

Manufacturing Experience

>
>
>
>

Emotional attachment:
DRAM = Intel

>

Only $100 million out of $1.6 billion of sales; Intel is already a logic
product company by 1985

Full Product Line

Late in 256K: 0.1% SOM

No full line since 1979
Down to 1 fab out of 8

Conclusion: Facts do not support beliefs. How do you explain those beliefs? Inertial forces in
corporate strategy: (1) bounded rationality and (2) emotional attachment.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 13

So, are DRAMs totally unimportant in 1984? No, because Sunlin Chou and the DRAM process
technology group is a key part of Intels distinctive technological competencies, which are
transferable to microprocessors. So, legitimate uncertainty about strategic importance of the
DRAM business is also a source of inertia in corporate strategy (must distinguish the product
market and distinctive competence aspects).

3. Why did Intel lose its leadership position in DRAMs? What has Intels technology
strategy been with respect to DRAMs? How has the organization reacted to the evolution of
the DRAM business? What does this say about Intels generic strategy and the image it has
of itself as a company? Why could the organization do so? What could Intel have done
differently? What would be the implications if it had done so? (Second middle board - 40
minutes)
Intel got into a precarious strategic position in the DRAM business because (i) industry evolution
caused DRAM to become a commodity business and (ii) because of its strategy in the DRAM
business, which continued to be based on its distinctive technological competence (IC design and
process technology; i.e., what Intel had got), while the basis of competition in the DRAM
industry (what it takes to win) had changed toward large scale precision manufacturing
engineering, and (iii) because of its internal selection environment, which had already diverted
critical manufacturing resources away from DRAM toward new specialty products. This case
offers clear evidence of inertia in technological competence deployment in the face of changing
basis of competition.
(i) Industry Evolution (Use Exhibit 1: The Extended Industry Analysis Framework)
* Competitors and new entrants:
- Mostek (4K, 16K): Introduce more user-friendly DRAM; but just another high-technology
startup, similar to Intel
- Japanese (64K): very different type of companies: large, vertically and horizontally
integrated; very strong in manufacturing; low capital cost; view DRAM as a strategic
component, so their intent is to dominate (use framework of Possible States Facing P)
* Government: Japanese (MITI) helps orchestrate the entry of Japanese companies into the
DRAM business
*Suppliers: Equipment suppliers more important for innovation; many now in Japan; work
together with their Japanese customers in different way than with US manufacturers; learn and
disseminate learning about process technology across their customers (harder to hold on to IP);
*Customers: Change from techies to purchasing departments, which requires far greater
emphasis on reliability, quality and price
*Technology: emergence of dominant design as of 4K product generation; technological
trajectory becomes highly predictable, which favors manufacturing competence; movement of
employees between companies (harder to hold on to IP)

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 14

Conclusion: DRAM has become a commodity business with attendant shifting basis of
competition. This is a classic example of P-independent change.
(ii) Intels DRAM Strategy (Use Exhibit 2: Forces Driving Company Evolution)
* 1K: Intel = technology leader based on process technology competence (fairly low yields, but
high enough to beat core memory standard product in the computer market segment)
*4K: Intel still technology leader, but Mostek enters with user-friendly design and dominant
design emerges
*16K: Intel focuses on innovative product: single-power supply; expects all vendors will have to
adopt this innovation and that 64K will be delayed. Unfortunately, single-power supply becomes
a niche product
*64K: Japanese enter with standard design on time; Intel is behind and tries to leapfrog Japanese
with redundancy, another design and process technology-based innovation. Redundancy does
not work well in large-scale manufacturing and Intel falls farther behind in the standard market
*256K: Intel is again late and now far behind the Japanese. Intel uses CMOS, a process
technology-based innovation, to try to catch up; CMOS is a critical advance, but remains a very
small niche product for this product generation
*1Meg: Intel now far behind and cannot hope to be technologically ahead across the board, so
chooses thin dielectrics, again a process technology-based advance, as the area of
differentiation. But now, strategic decisions about capacity investments become paramount. Top
management decides not to make these investments.
Conclusion: Intel continues to try to compete using its process technology competence (what it
has got) as if the DRAM business was still a specialty business (what it takes to win) and to
leapfrog the competition in the face of DRAM commoditization. Intel competes with wrong
weapon; strong inertia in competence deployment. Note: The instructor can refer here to other
studies that have shown strong inertial forces in competence deployment (e.g.. Cooper and
Schendel, Business Horizons, 1976; Henderson and Clark, ASQ, 1990).
(iii) Intels Internal selection Environment (Use Exhibit 2: Dynamic Forces Driving Company
Evolution, and also introduce ideas from Exhibit 3: A Framework of the Strategy-Making
Process in Established Companies)
* Manufacturing resource allocation: The DRAM competed with the EPROM and logic
businesses whenever manufacturing capacity was scarce. Intels internal selection environment
played a key role in motivating middle-level managers to shift scarce manufacturing resources to
the new, higher-margin businesses, even though the official corporate strategy was still stated in
terms of Intel the memory company. The shift took place because Intel middle-level managers
used the maximize margin-per-waferstart rule to allocate fab capacity. Other middle managers
(Ron Smith) also made crucial technology choices (discontinue 4-transistor SRAM process) that
signaled a belief that microprocessors were more important than memories for Intels future.
Note that the maximize margin-per-waferstart rule was introduced by top management and
reflects Intels generic strategy of product leader/differentiator. This rule also reflects Intels

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 15

culture. Top management did not anticipate that this rule would get the company out of its core
memory business. Yet when it did, top management found it did not want to change it. They
probably realized that doing so would probably have created tremendous confusion within the
company and lead to commitments that they were not ready to make. Of course, each of the
marginal allocation decisions were not viewed as strategic, but the sum of all these incremental
decisions surely was.
R&D resource allocation: At the same time, top management continued to spend one-third of
available R&D resources to the DRAM business (case exhibit 6). But by the end of 1984, there is
a major discrepancy between the fab resource allocation and the R&D resource allocation, which
needs to be addressed by top management. They will have to change one or the other in order to
resolve the disharmony and attendant strategic dissonance that now has become very manifest in
the organization. Here the instructor can draw a figure that shows (on the y-axis) the declining
DRAM fab allocation over time (x-axis) and juxtaposes it to the DRAM R&D allocation (also on
the y-axis), which remains basically constant over time (see below).
Resources
R&D allocation (top mgt)

Mfg capacity allocation (middle mgt)

1984
Time
This is a powerful way to illustrate graphically the enormous strategic dissonance that must
have existed at the time.
Conclusion: Official corporate strategy (what we say) has significantly diverged from strategic
action (what we do). The internal selection environment, reflecting the competitive reality faced
by DRAM and Intels other products, has already moved the company out of DRAM. Intels
generic strategy (differentiation and technological leadership) shows greater robustness than its
substantive strategy (which particular products to bring to market), but this was only possible
because the companys distinctive competence has spawned new business opportunities that
were viable in the external selection environment and could be supported by the internal
selection environment.
Note that it could be argued that Intel should have tried to match the changing basis of
competition (importance of large-scale precision manufacturing competence) and try to compete
effectively with the Japanese new entrants. However, in that case, and assuming that Intel could

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 16

have developed such competence sufficiently quickly (a strong assumption), the company would

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 17

probably have been an also ran in a highly cyclical commodity business. This would also have
involved a major cultural change with hard to predict consequences.
4. How did Intel get into the microprocessor business? How important is this business for
Intel at the end of 1984? (Second sideboard - 15 minutes)
* The discussion should bring out that microprocessors (like EPROMs) were not a strategically
planned product, but came about as the result of the efforts of a key technologist (Ted Hoff) in
response to a customer request for a specialized chip set (a Japanese company that wanted to
bring out an electronic desktop calculator and needed integrated circuitry). Here Exhibit 3: A
Framework of the Strategy-Making Process in Established Companies is helpful to illuminate
the fact that this project was clearly an autonomous one relative to the official corporate
strategy at the time. Note that Intel had sold the rights to what became the microprocessor for
$60,000 to the Japanese customer! This clearly indicates that Intel top management had not
strategically planned this product, nor that they initially realized its strategic importance
beyond the fact that it constituted a major technological breakthrough (but what were its
applications?). Fortunately, Hoff and others were quickly able to see various vertical market
segments that could be served by the new product (mostly embedded controller applications).
* This segment of the discussion should also bring out that microprocessors was a specialty
business in which differentiation was still possible based on Intels distinctive competencies in
IC design and process technology, which were widely recognized in the industry and explain
why the product development engineers of the Japanese company went to Intel to procure the
chips they needed for their electronic desktop calculator products.
* In 1984 Intels core business is already logic (microprocessor) products (case exhibit 7). The
instructor can translate this exhibit in a graph, shown in Exhibit 4: that shows the declining
memory business and the increasing logic business over time, with the curves crossing around
1981-82
* Top managements role was in recognizing the strategic importance of the new device and in
supporting the development of Intels design competencies before the market signals on the
importance of the new device were very clear, and in ratifying the decisions made by middlelevel managers. This discussion offers the opportunity to examine the differences between
strategic management based on vision and strategic recognition (see additional readings at
the end of this teaching note).
5. What is the strategic conundrum that Intel top management has faced since 1982, and
how can it be resolved by Dr. Grove at the end of 1984 in order to drive Intel in a viable
new strategic direction? (Second sideboard - 10 minutes)
* Top management was quite slow in exiting from the DRAM business and in declaring formally
that Intel had become a microprocessor company. One reason for the slowness - a remarkable
fact given the generally recognized superior intellectual and executive prowess of persons such
as Gordon Moore and Andy Grove was the intertwining of the companys best process
technology resource (Sunlin Chou and his group) with the losing DRAM business.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 18

* Another factor was the strategic conundrum in which top management found itself during the
period late-1970s till 1985. This conundrum involved the fact that Intel was losing its strategic
position in the very large and fast growing DRAM business, while simultaneously achieving
success with its novel logic products in a variety of vertical market segments (mostly
embedded applications). Top management realized, however, that the sum of a bunch of niche
markets was not equivalent to a big market and thus was reluctant to let go of the DRAM
business.
* Intel was in a sense lucky that the new PC market segment began to take off rapidly in the
early 1980s. Andy Grove then realized that the PC market segment could be the new big
market segment for Intel and was willing to bet on this opportunity. Once he came to that
conclusion, it made sense to give up on the other big market segment (DRAM) in which Intels
defeat was all but definitive.
UPDATE
In November 1984, Intel decided not to put the 1 MB DRAM in production. The main reason
was that they did not want to invest the several hundred million dollars in new plant and
equipment that it would have taken to try to compete effectively with the Japanese. This was a
defacto exit decision, but the implementation lingered on throughout the first three quarters of
1985. The DRAM process technology group was allowed to continue work on prototypes for the
1 MB DRAM. They produced working prototypes in March 1985. The general manager of the
Components division to whom the DRAM business reported continued to try to find ways
to stay in DRAM looking for manufacturing partners. In summer 1985 Andy Grove reassigned
this manager and personally took charge of the implementation of the exit decision. The process
technology group for microprocessors, based in Livermore, California, was transferred to Oregon
and combined with the DRAM process technology group. Sunlin Chous group was given the
assignment to develop the 1 micron version of the new 386 microprocessor. In October 1985, the
last fab still manufacturing 64K and 256K DRAMs was closed. Several hundred manufacturing
personnel were laid off. Some DRAM designers left the company. Andy Grove went to Oregon
and gave a talk to the process technology group saying Welcome to the mainstream Intel,
thereby explicitly stating Intels new corporate strategy as a microcomputer company.
SUMMARY
1. Five dynamic forces drive firm evolution. In the state of harmony (1) distinctive competence
is consistent with (2) the basis of competition in the industry; (3) official corporate strategy and
(4) strategic action are aligned; and (5) the internal selection environment is relatively peaceful
and remains in the background.
2. Sooner or later, industries undergo major changes causing a shift in the basis of competition
and disharmony among the forces. Such P-independent change, in turn, creates an inflection
point for the firm and causes strategic dissonance in the organization, which raises the
importance of the internal selection environment.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 19

3. Strategic recognition of inflection points is difficult. In the face of changes in the basis of
competitive advantage, firms tend to evoke inertial responses in competence deployment and in
corporate strategy.
4. These inertial forces are associated with the perceived relative importance of different
competences in light of the firms history (e.g., process technology at Intel) and with the image
that top management has about the firms identity (Intel the memory company).
5. Exiting from a business is not just a product-market (position) decision, because key elements
of distinctive competence may be tied up with that (losing) business. The difficulty of sorting out
the implications of exit for the firms distinctive competence versus for its product-market
position may delay the strategic exit decision.
6. Strategic dissonance becomes manifest as discrepancies emerge between official corporate
strategy (what we say) and strategic action (what we do) and/or between the basis of competition
(what it takes to win) and distinctive competence (what weve got).
7. In dynamic industries, a firms survival may depend more on the capacity of its internal
selection environment to make strategic action respond to the reality of the external environment
than on making strategic action consistent with stated strategy.
8. Technological competencies sometimes generate new product-market opportunities that are
outside the scope of the official corporate strategy. Taking advantage of these is a key strategic
management challenge.
9. A firms product-market activities change over time. This happens, in part, by top
management design but also as the outcome of internal resource allocation processes. In reality,
corporate strategy is what the firm does, not what top management says. Eventually, of course,
strategy and action must be brought in line.
10. But bringing strategy and action in line is difficult as long as top management does not have
a clear alternative for a losing core business. Intel top management was able to make the difficult
decision to exit from DRAM (a very large market) only after the emergence of the PC market (a
new very large market) provided a viable alternative. Until then, top management faced the
conundrum of giving up a very large market segment (DRAM) for a bunch of relatively small
niche market segments for its logic products. They realized that the sum of a bunch of niches is
not equal to a large market!
ADDITIONAL READING
Burgelman, R.A. Intraorganizational Ecology of Strategy-Making and Organizational
Adaptation: Theory and Field Research, Organization Science, August 1991
Burgelman, R.A. Fading Memories: A Process Theory of Strategic Business Exit in Dynamic
Environments. Administrative Science Quarterly, March 1994.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 20

Burgelman, R.A. A Process Model of Strategic Business Exit: Implications for an Evolutionary
perspective on Strategy, Strategic Management Journal, Special Issue, September 1996.
Burgelman, R.A. Strategy is Destiny: How Strategy-making Shapes a Companys Future, New
York, Free Press, 2002, chapters 2-5.
Burgelman, R.A. and Grove, A.S., Strategic Dissonance, California Management Review,
Winter 1996.
Grove, A/S. Only the Paranoid Survive, New York, DoubleDay, 1996.

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

Exhibit 1: The Extended Industry Analysis Framework

Adapted 7 Forces Model
(* indicates forces added by Grove)
Complementors*
Competitors
Channels*

The Business
Customers

New
Entrants

Substitutes

Source: Grove, A.S., Lecture material, mid-1990s

Suppliers

p. 21

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 22

Exhibit 2 Dynamic Forces Driving Company Evolution

BASIS OF COMPETITIVE
ADVANTAGE INTHE
INDUSTRY

OFFICIAL
CORPORATE
STRATEGY

INTERNAL SELECTION
ENVIRONMENT

STRATEGIC
ACTION

DISTINCTIVECOMPETENCE
OF THEFIRM

Source: Burgelman, R.A., Fading Memories: A Process Theory of Strategic Business Exit in
Dynamic Environments, Administrative Sciences Quarterly, March 1994, p. 31

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 23

Exhibit 3: A Framework of the Strategy-Making Process in Established Companies

Autonomous
Strategic Action

Strategic
Context

Concept of
Corporate

Induced Strategic
Action

Structural
Context

Strategy

Source: Burgelman, R.A., Strategy is Destiny: How Strategy-Making Shapes a Companys

Future, New York, The Free Press, 2002, p. 9

Teaching Note for Intel Corporation (A): The DRAM Decision BP-256 TN

p. 24

Exhibit 4: Intels Evolution form Memory Company to Microcomputer Company

Percentage of Intels
Revenues

o
Re

ion
tat
n
rie

Logic

Microprocessor company

Memory Company

Memories

1972

1973 1974

1975 1976 1977

1978

1979 1980 1981 1982

1983 1984 1984

1986

1987

1988

Year

Intel: The LSI memory supplier

Intel: The microprocessor company

Source: Burgelman, R.A., Strategy is Destiny: How Strategy-Making Shapes a Companys

Future, New York, The Free Press, 2002.